Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 208079 - [feature-request] app-portage/autounmask -- please don't add redundant ~arch to package.keywords
Summary: [feature-request] app-portage/autounmask -- please don't add redundant ~arch ...
Status: RESOLVED WONTFIX
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: New packages (show other bugs)
Hardware: All Linux
: High enhancement
Assignee: Christian Hartmann (RETIRED)
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-01-29 18:53 UTC by Steve L
Modified: 2008-09-10 12:30 UTC (History)
0 users

See Also:
Package list:
Runtime testing required: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Steve L 2008-01-29 18:53:59 UTC
Eg app-foo/baz ~amd64
on an amd64 machine just looks untidy, since we've been recommended for ages simply to use the cat/pkg-name (+ ver if we want it) and leave off default arch.

It makes it trickier to spot unusual settings.
Comment 1 Jakub Moc (RETIRED) gentoo-dev 2008-01-29 21:07:46 UTC
(In reply to comment #0)
> on an amd64 machine just looks untidy, since we've been recommended for ages
> simply to use the cat/pkg-name (+ ver if we want it) and leave off default
> arch.

I don't know who's recommended that to you but it's plain wrong; explicit ACCEPT_KEYWORDS is not a bug.
Comment 2 Steve L 2008-01-30 00:20:25 UTC
Jakub-- I never said it was incorrect. That's why this was an enhancement request.

The point is the architecture specification there is redundant. The exact same masking as:
cat-foo/pkg-bar ~amd64
is achieved with:
cat-foo/pkg-bar
in package.keywords on an amd64 machine. All I want is for the script to do what a user would do, and thus keep the file as cruft-free as possible.

Don't get me wrong: this is a minor enhancement, but it would make autounmask just about perfect imo. It's not an actual bug in the functionality.
Comment 3 Jakub Moc (RETIRED) gentoo-dev 2008-01-30 00:30:35 UTC
Eh, I really fail to see how specifying accepted keywords counts as 'cruft'. The 'implicit' behaviour doesn't make sense and should be nuked from portage.
Comment 4 Steve L 2008-02-12 19:14:07 UTC
(In reply to comment #3)
> Eh, I really fail to see how specifying accepted keywords counts as 'cruft'.
If it's redundant, it's cruft. If I wanted an x86-only package it would make sense for me to specify x86 or ~x86 as required.

> The 'implicit' behaviour doesn't make sense and should be nuked from portage.
>
Eh? It makes perfect sense: unstable on current architecture. This makes it far easier to copy a world file to another machine, as well as being cleaner.

Why do you say it should be nuked? It's hardly causing user confusion.. I look forward to the GLEP :p
Comment 5 Christian Hartmann (RETIRED) gentoo-dev 2008-02-21 18:39:55 UTC
Hmm. I could make this an option but I'm not sure if it's handy for the user to have just another parameter to type in every time. A configuration file also seems overkill to me.

What's your thought on this Steve?
Comment 6 Steve L 2008-02-22 13:36:11 UTC
(In reply to comment #5)
> Hmm. I could make this an option but I'm not sure if it's handy for the user to
> have just another parameter to type in every time. A configuration file also
> seems overkill to me.
> 
> What's your thought on this Steve?
> 
Hi Ian; I agree it doesn't need a config file or another option. I was hoping it would just be the default behaviour, tbh. (I use autounmask a lot nowadays, since it's called by update.) Is there a reason why ~arch would be preferred?

I can deal with another option since my use of it is scripted (though I need to consider how we call it, since we always just use -n which might not be the best thing) if that's what you settle on (and don't mind adding it, ofc.)

Thanks for the great app though; I noticed it deleted some old entries, which impressed me no end.
Comment 7 Christian Hartmann (RETIRED) gentoo-dev 2008-09-10 12:30:33 UTC
Retiring..